--- jupytext: text_representation: extension: .md format_name: myst format_version: 0.12 jupytext_version: 1.9.1 kernelspec: display_name: Python 3 language: python name: python3 --- :::{currentmodule} tskit ::: (sec_data_model)= # Data model The `tskit` library deals with sets of sampled genome sequences through storage and analysis of their shared genetic ancestry. This genealogical ancestry (sometimes known as an Ancestral Recombination Graph) is stored concisely in `tskit` in the "succinct tree sequence" format, which comprises a collection of easy-to-understand tables. This page documents the structure of the tables and encoding of table data, as well as the encoding of the correlated genetic trees that can be extracted from a `tskit` tree sequence. We begin by defining the the structure of the tables in the {ref}`sec_table_definitions` section. The {ref}`sec_data_model_data_encoding` section then describe how data is stored in those tables (also see the {ref}`sec_file_formats` chapter). The {ref}`sec_data_model_tree_structure` section then describes the encoding of the trees that are generated from the {class}`NodeTable` and {class}`EdgeTable`. Finally, we describe how genotype data arises from tree structure, especially how we can incorporate the idea of missing data. (sec_table_definitions)= ## Table definitions (sec_table_types_definitions)= ### Table types A tree sequence can be stored in a collection of eight tables: {ref}`Node `, {ref}`Edge `, {ref}`Individual `, {ref}`Site `, {ref}`Mutation `, {ref}`Migration `, {ref}`Population `, and {ref}`Provenance `. The Node and Edge tables store the genealogical relationships that define the trees, and the Individual table describes how multiple genomes are grouped within individuals; the Site and Mutation tables describe where mutations fall on the trees; the Migration table describes how lineages move across space; and the Provenance table contains information on where the data came from. Only Node and Edge tables are necessary to encode the genealogical trees; Sites and Mutations are optional but necessary to encode polymorphism (sequence) data; the remainder are optional. In the following sections we define these components of a tree sequence in more detail. (sec_node_table_definition)= #### Node Table A **node** defines a monoploid set of chromosomes (a "genome") of a specific individual that was born at some time in the past: the set of chromosomes inherited from a particular one of the individual's parents. (See {ref}`sec_nodes_or_individuals` for more discussion.) Every vertex in the marginal trees of a tree sequence corresponds to exactly one node, and a node may be present in many trees. The node table contains five columns, of which `flags` and `time` are mandatory: | Column | Type | Description | | :------------ | ----------- | -------------------------------------: | | flags | uint32 | Bitwise flags. | | time | double | Birth time of node. | | population | int32 | Birth population of node. | | individual | int32 | The individual the node belongs to. | | metadata | binary | Node {ref}`sec_metadata_definition`. | The `time` column records the birth time of the individual in question, and is a floating point value. Similarly, the `population` column records the ID of the population where this individual was born. If not provided, `population` defaults to the null ID (-1). Otherwise, the population ID must refer to a row in the {ref}`sec_population_table_definition`. The `individual` column records the ID of the {ref}`Individual ` individual that this node belongs to. If specified, the ID must refer to a valid individual. If not provided, `individual` defaults to the null ID (-1). The `flags` column stores information about a particular node, and is composed of 32 bitwise boolean values. Currently, the only flag defined is `NODE_IS_SAMPLE = 1`, which defines the *sample* status of nodes. Marking a particular node as a "sample" means, for example, that the mutational state of the node will be included in the genotypes produced by {meth}`TreeSequence.variants`. Bits 0-15 (inclusive) of the `flags` column are reserved for internal use by `tskit` and should not be used by applications for anything other than the purposes documented here. Bits 16-31 (inclusive) are free for applications to use for any purpose and will not be altered or interpreteted by `tskit`. See the {ref}`sec_node_requirements` section for details on the properties required for a valid set of nodes. For convenience, the {ref}`text format ` for nodes decomposes the `flags` value into its separate values. Thus, in the text format we have a column for `is_sample`, which corresponds to the `flags` column in the underlying table. As more flags values are defined, these will be added to the text file format. The `metadata` column provides a location for client code to store information about each node. See the {ref}`sec_metadata_definition` section for more details on how metadata columns should be used. :::{note} The distinction between `flags` and `metadata` is that flags holds information about a node that the library understands, whereas metadata holds information about a node that the library *does not* understand. Metadata is for storing auxiliarly information that is not necessary for the core tree sequence algorithms. ::: (sec_individual_table_definition)= #### Individual Table An **individual** defines how nodes (which can be seen as representing single chromosomes) group together in a polyploid individual. The individual table contains three columns, of which only `flags` is mandatory. | Column | Type | Description | | :------------ | ---------- | -----------------------------------------: | | flags | uint32 | Bitwise flags. | | location | double | Location in arbitrary dimensions. | | parents | int32 | Ids of parent individuals. | | metadata | binary | Individual {ref}`sec_metadata_definition`. | See the {ref}`sec_individual_requirements` section for details on the properties required for a valid set of individuals. The `flags` column stores information about a particular individual, and is composed of 32 bitwise boolean values. Currently, no flags are defined. Bits 0-15 (inclusive) of the `flags` column are reserved for internal use by `tskit` and should not be used by applications for anything other than the purposes documented here. Bits 16-31 (inclusive) are free for applications to use for any purpose and will not be altered or interpreteted by `tskit`. The `location` column stores the location of an individual in arbitrary dimensions. This column is {ref}`ragged `, and so different individuals can have locations with different dimensions (i.e., one individual may have location `[]` and another `[0, 1, 0]`. This could therefore be used to store other quantities (e.g., phenotype). The `parents` column stores the ids of other individuals that are the parents of an individual. This can be used to store pedigree information for individuals. This column is {ref}`ragged ` such that an individual can have any number of parents. The `metadata` column provides a location for client code to store information about each individual. See the {ref}`sec_metadata_definition` section for more details on how metadata columns should be used. :::{note} The distinction between `flags` and `metadata` is that flags holds information about a individual that the library understands, whereas metadata holds information about a individual that the library *does not* understand. Metadata is for storing auxiliarly information that is not necessary for the core tree sequence algorithms. ::: (sec_edge_table_definition)= #### Edge Table An **edge** defines a parent-child relationship between a pair of nodes over a specific sequence interval. The edge table contains five columns, all of which are mandatory except `metadata`: | Column | Type | Description | | :------------ | ---------- | -----------------------------------------: | | left | double | Left coordinate of the edge (inclusive). | | right | double | Right coordinate of the edge (exclusive). | | parent | int32 | Parent node ID. | | child | int32 | Child node ID. | | metadata | binary | Node {ref}`sec_metadata_definition`. | Each row in an edge table describes a half-open genomic interval `[left, right)` over which the `child` inherited from the given `parent`. The `left` and `right` columns are defined using double precision floating point values. The `parent` and `child` columns specify integer IDs in the associated {ref}`sec_node_table_definition`. The `metadata` column provides a location for client code to store information about each edge. See the {ref}`sec_metadata_definition` section for more details on how metadata columns should be used. See the {ref}`sec_edge_requirements` section for details on the properties required for a valid set of edges. (sec_site_table_definition)= #### Site Table A **site** defines a particular location along the genome in which we are interested in observing the allelic state. The site table contains three columns, of which `position` and `ancestral_state` are mandatory. | Column | Type | Description | | :-------------- | ---------- | -----------------------------------------: | | position | double | Position of site in genome coordinates. | | ancestral_state | text | The state at the root of the tree. | | metadata | binary | Site {ref}`sec_metadata_definition`. | The `position` column is a floating point value defining the location of the site in question along the genome. The `ancestral_state` column specifies the allelic state at the root of the tree, thus defining the state that nodes inherit if no mutations intervene. The column stores text character data of arbitrary length. The `metadata` column provides a location for client code to store information about each site. See the {ref}`sec_metadata_definition` section for more details on how metadata columns should be used. See the {ref}`sec_site_requirements` section for details on the properties required for a valid set of sites. (sec_mutation_table_definition)= #### Mutation Table A **mutation** defines a change of allelic state on a tree at a particular site. The mutation table contains five columns, of which `site`, `node` and `derived_state` are mandatory. | Column | Type | Description | | :-------------- | ---------- | ---------------------------------------------: | | site | int32 | The ID of the site the mutation occurs at. | | node | int32 | The node this mutation occurs at. | | parent | int32 | The ID of the parent mutation. | | time | double | Time at which the mutation occurred. | | derived_state | char | The allelic state resulting from the mutation. | | metadata | binary | Mutation {ref}`sec_metadata_definition`. | The `site` column is an integer value defining the ID of the {ref}`site ` at which this mutation occurred. The `node` column is an integer value defining the ID of the first {ref}`node ` in the tree below this mutation. The `time` column is a double precision floating point value recording how long ago the mutation happened. The `derived_state` column specifies the allelic state resulting from the mutation, thus defining the state that the `node` and any descendant nodes in the subtree inherit unless further mutations occur. The column stores text character data of arbitrary length. The `parent` column is an integer value defining the ID of the mutation whose allelic state this mutation replaced. If there is no mutation at the site in question on the path back to root, then this field is set to the null ID (-1). (The `parent` column is only required in situations where there are multiple mutations at a given site. For "infinite sites" mutations, it can be ignored.) The `metadata` column provides a location for client code to store information about each site. See the {ref}`sec_metadata_definition` section for more details on how metadata columns should be used. See the {ref}`sec_mutation_requirements` section for details on the properties required for a valid set of mutations. (sec_migration_table_definition)= #### Migration Table :::{note} Encoding migration in the migrations table is a legacy approach associated with older versions of `msprime`; recording movement between populations in the migration table is entirely optional, even when related nodes are assigned to different populations. ::: :::{warning} The migration table may be entirely removed from the `tskit` data model in the future. Meanwhile, a number of `tskit` functions, such as {meth}`~TreeSequence.simplify()` will raise an error if data exists in the migrations table. ::: :::{seealso} The {ref}`msprime:sec_ancestry_record_migrations` sections and the associated discussion of {ref}`msprime:sec_demography_migration` in the `msprime` documentation. ::: In simulations, trees can be thought of as spread across space, and it is helpful for inferring demographic history to record this history. Migrations are performed by individual ancestors, but might not be tagged by an individual whose genome is tracked as a `node` (as in a discrete-deme model they are unlikely to be both a migrant and a most recent common ancestor). So, `tskit` can record separately when a segment of ancestry has moved between populations. This table is not required, even if different nodes come from different populations. | Column | Type | Description | | :--------- | -------- | -----------------------------------------------------: | | left | double | Left coordinate of the migrating segment (inclusive). | | right | double | Right coordinate of the migrating segment (exclusive). | | node | int32 | Node ID. | | source | int32 | Source population ID. | | dest | int32 | Destination population ID. | | time | double | Time of migration event. | | metadata | binary | Migration {ref}`sec_metadata_definition`. | The `left` and `right` columns are floating point values defining the half-open segment of genome affected (these need not exactly correspond to breakpoints between edges). The `source` and `dest` columns record the IDs of the respective populations (note that by `msprime` convention, "source" and "destination" are defined in reverse time, see {ref}`msprime:sec_demography_direction_of_time`.). The `time` column holds floating point values recording the time of the event, with migrations assumed to occur instantaneously. The `node` column records the ID of the child node of the migrating segment; in consequence the population ID of the `node` will match the `src` ID (unless sequential migrations affect the same `node`, in which case it will match the `src` value of the youngest of those migrations). The `metadata` column provides a location for client code to store information about each migration. See the {ref}`sec_metadata_definition` section for more details on how metadata columns should be used. See the {ref}`sec_migration_requirements` section for details on the properties required for a valid set of migrations. (sec_population_table_definition)= #### Population Table A **population** defines a grouping of individuals that a node can be said to belong to. The population table contains one column, `metadata`. | Column | Type | Description | | :--------- | -------- | -----------------------------------------: | | metadata | binary | Population {ref}`sec_metadata_definition`. | The `metadata` column provides a location for client code to store information about each population. See the {ref}`sec_metadata_definition` section for more details on how metadata columns should be used. See the {ref}`sec_population_requirements` section for details on the properties required for a valid set of populations. (sec_provenance_table_definition)= #### Provenance Table | Column | Type | Description | | :-------- | ----- | ----------------------------------------------------------------------: | | timestamp | char | Timestamp in [ISO-8601](https://en.wikipedia.org/wiki/ISO_8601) format. | | record | char | Provenance record as JSON. | (sec_metadata_definition)= ### Metadata Each table (excluding provenance) has a metadata column for storing and passing along information that tskit does not use or interpret. See {ref}`sec_metadata` for details. The metadata columns are {ref}`binary columns `. When using the {ref}`sec_text_file_format`, metadata values are written as opaque text. By default, :meth:`TreeSequence.dump_text` will base64-encode metadata values that are stored as raw bytes (when ``base64_metadata=True``) so that binary data can be safely printed and exchanged; in this case :func:`tskit.load_text` will base64-decode the corresponding text fields back to bytes. When metadata has already been decoded to a structured Python object (for example via a metadata schema), the textual representation written by :meth:`TreeSequence.dump_text` is the ``repr`` of that object, and :func:`tskit.load_text` does not attempt to reconstruct the original structured value from this representation. For reliable metadata round-tripping, prefer the native binary tree sequence file format over the text formats. The tree sequence itself also has metadata stored as a byte array. (sec_valid_tree_sequence_requirements)= ### Valid tree sequence requirements Arbitrary data can be stored in tables using the classes in the {ref}`sec_tables_api`. The {meth}`TableCollection.tree_sequence` method can be used to turn such a {class}`TableCollection` into an immutable {class}`TreeSequence` object, but this requires the tables to fulfil a specific set of requirements. In this section we list these requirements, and explain their rationale. Violations of most of these requirements are detected when the user attempts to load a tree sequence via {func}`tskit.load` or {meth}`TableCollection.tree_sequence`, raising an informative error message. Some more complex requirements may not be detectable at load-time, and errors may not occur until certain operations are attempted. These are documented below. At the tree-sequence level, we require that the coordinate space has a finite, strictly positive length; that is, the `sequence_length` attribute must be a finite value greater than zero. The Python API also provides tools that can transform a collection of tables into a valid collection of tables, so long as they are logically consistent, see {ref}`sec_tables_api_creating_valid_tree_sequence`. (sec_individual_requirements)= #### Individual requirements Individuals are a basic type in a tree sequence and are not defined with respect to any other tables. Individuals can have a reference to their parent individuals, if present these references must be valid or null (-1). An individual cannot list itself as its own parent. A valid tree sequence does not require individuals to be sorted in any particular order, and sorting a set of tables using {meth}`TableCollection.sort` has no effect on individuals. However, individuals can be optionally sorted using {meth}`TableCollection.sort_individuals`. (sec_node_requirements)= #### Node requirements Given a valid set of individuals and populations, the requirements for each node are: - `time` must be a finite (non-NaN, non-infinite) value; - `population` must either be null (-1) or refer to a valid population ID; - `individual` must either be null (-1) or refer to a valid individual ID. An ID refers to a zero-indexed row number in the relevant table, and so is "valid" if is between 0 and one less than the number of rows in the relevant table. There are no requirements regarding the ordering of nodes with respect to time. Sorting a set of tables using {meth}`TableCollection.sort` has no effect on nodes. (sec_edge_requirements)= #### Edge requirements Given a valid set of nodes and a sequence length {math}`L`, the simple requirements for each edge are: - We must have finite coordinates with {math}`0 \leq` `left` {math}`<` `right` {math}`\leq L`; - `parent` and `child` must be valid node IDs; - `time[parent]` > `time[child]`; - edges must be unique (i.e., no duplicate edges are allowed). The first requirement simply ensures that the interval makes sense. The third requirement ensures that we cannot have loops, since time is always increasing as we ascend the tree. To ensure a valid tree sequence there is one further requirement: - The set of intervals on which each node is a child must be disjoint. This guarantees that we cannot have contradictory edges (i.e., where a node `a` is a child of both `b` and `c`), and ensures that at each point on the sequence we have a well-formed forest of trees. In the interest of algorithmic efficiency, edges must have the following sortedness properties: - All edges for a given parent must be contiguous; - Edges must be listed in nondecreasing order of `parent` time; - Within the edges for a given `parent`, edges must be sorted first by `child` ID and then by `left` coordinate. Violations of these requirements are detected at load time. The {meth}`TableCollection.sort` method will ensure that these sortedness properties are fulfilled. (sec_site_requirements)= #### Site requirements Given a valid set of nodes and a sequence length {math}`L`, the simple requirements for a valid set of sites are: - We must have a finite coordinate with {math}`0 \leq` `position` {math}`< L`; - `position` values must be unique. For simplicity and algorithmic efficiency, sites must also: - Be sorted in increasing order of `position`. Violations of these requirements are detected at load time. The {meth}`TableCollection.sort` method ensures that sites are sorted according to these criteria. (sec_mutation_requirements)= #### Mutation requirements Given a valid set of nodes, edges and sites, the requirements for a valid set of mutations are: - `site` must refer to a valid site ID; - `node` must refer to a valid node ID; - `time` must either be `UNKNOWN_TIME` (a NAN value which indicates the time is unknown) or be a finite value which is greater or equal to the mutation `node`'s `time`, less than the `node` above the mutation's `time` and equal to or less than the `time` of the `parent` mutation if this mutation has one. If one mutation on a site has `UNKNOWN_TIME` then all mutations at that site must, as a mixture of known and unknown is not valid. - `parent` must either be the null ID (-1), if the mutation has no parent, or a valid mutation ID within the current table. Furthermore, - The `parent` value must be consistent with the topology of the tree at the site of the mutation, such that a path from the child mutation to the parent mutation exists without passing through any other mutations at the same site. For simplicity and algorithmic efficiency, mutations must also: - be sorted by site ID; - when there are multiple mutations per site, mutations should be ordered by decreasing time, if known, and parent mutations must occur **before** their children (i.e. if a mutation with ID {math}`x` has `parent` with ID {math}`y`, then we must have {math}`y < x`). Violations of these sorting requirements are detected at load time. The {meth}`TableCollection.sort` method ensures that mutations are sorted according site ID, but does not at present enforce that mutations occur after their parent mutations. Silent mutations (i.e., mutations for which the ancestral and derived states are the same) are allowed. For example, if we have a site with ancestral state of "A" and a single mutation with derived state "A", then this mutation does not result in any change of state. (This addition was made in release C_0.99.11.) :::{note} As `tskit.UNKNOWN_TIME` is implemented as a `NaN` value, tests for equality will always fail. Use `tskit.is_unknown_time` to detect unknown values. ::: (sec_migration_requirements)= #### Migration requirements Given a valid set of nodes and edges, the requirements for a valid set of migrations are: - `left` and `right` must be finite values that lie within the tree sequence coordinate space (i.e., from 0 to `sequence_length`), with {math}`0 \leq` `left` {math}`<` `right` {math}`\leq L`; - `node` must be a valid node ID; - if population references are checked, `source` and `dest` must be valid population IDs; - `time` must be a finite value. To enable efficient processing, migrations must also be sorted by nondecreasing `time` value. Conceptually, a migration records that a segment of ancestry for the given `node` moves between populations along the tree. In typical demographic models we expect: - `time` to lie strictly between the time of the migrating `node` and the time of any ancestral node from which that node inherits on the segment `[left, right)`; - the `population` of any such ancestor to match the `source` population, until another `migration` intervenes. These conceptual relationships are not currently validated. It is the responsibility of code that creates migrations to satisfy them where required. Note in particular that there is no requirement that adjacent migration records should be "squashed". That is, we can have two records `m1` and `m2` such that `m1.right` = `m2.left` and with the `node`, `source`, `dest` and `time` fields equal. This is because such records will usually represent two independent ancestral segments migrating at the same time, and as such squashing them into a single record would result in a loss of information. (sec_population_requirements)= #### Population requirements There are no requirements on a population table. (sec_provenance_requirements)= #### Provenance requirements The `timestamp` column of a provenance table should be in [ISO-8601](https://en.wikipedia.org/wiki/ISO_8601) format. The `record` column stores a JSON document describing how and where the tree sequence was produced. For tree sequences generated by tskit and related tools, this JSON is expected to conform to the :ref:`provenance schema ` described in {ref}`sec_provenance`. (sec_table_indexes)= ### Table indexes To efficiently iterate over the trees in a tree sequence, `tskit` uses indexes built on the edges. To create a tree sequence from a table collection the tables must be indexed; the {meth}`TableCollection.build_index` method can be used to create an index on a table collection if necessary. :::{todo} Add more details on what the indexes actually are. ::: (sec_data_model_saving)= ### Saving to file When serializing (e.g. storing a {class}`TreeSequence` to disk using {meth}`dump`), the underlying tables are stored along with the indexes, top-level metadata, attributes such as the sequence length and time units, and the {ref}`sec_data_model_reference_sequence` if it exists. {func}`Loading ` such a file returns an immutable tree sequence object, with pre-calculated indexes immediately available. See the {ref}`sec_tree_sequence_file_format` section for more details. Although data in a raw {class}`TableCollection` need not conform to the {ref}`sec_valid_tree_sequence_requirements`, it too can be {meth}`dumped ` to a file (with indexes stored if they exist). (sec_data_model_data_encoding)= ## Data encoding In this section we describe the high-level details of how data is encoded in tables. Tables store data in a **columnar** manner. In memory, each table is organised as a number of blocks of contiguous storage, one for each column. There are many advantages to this approach, but the key property for us is that allows for very efficient transfer of data in and out of tables. Rather than inserting data into tables row-by-row (which can be done in Python {ref}`using the add_row methods`), it is much more efficient to add many rows at the same time by providing pointers to blocks of contiguous memory. By taking this approach, we can work with tables containing gigabytes of data very efficiently. For instance, in the {ref}`sec_python_api` we can use the [numpy Array API](https://docs.scipy.org/doc/numpy-1.13.0/reference/arrays.html) to allow us to define and work with numeric arrays of the required types. Node IDs, for example, are defined using 32 bit integers. Thus, the `parent` column of an {ref}`sec_edge_table_definition`'s with `n` rows is a block `4n` bytes. This approach is very straightforward for columns in which each row contains a fixed number of values. However, dealing with columns containing a **variable** number of values is more problematic. (sec_encoding_ragged_columns)= ### Encoding ragged columns A **ragged** column is a column in which the rows are not of a fixed length. For example, {ref}`sec_metadata_definition` columns contain binary of data of arbitrary length. To encode such columns in the tables API, we store **two** columns: one contains the flattened array of data and another stores the **offsets** of each row into this flattened array. Consider an example: ```{code-cell} ipython3 import tskit s = tskit.SiteTable() s.add_row(0, "A") s.add_row(0, "") s.add_row(0, "TTT") s.add_row(0, "G") s ``` In this example we create a {ref}`sec_site_table_definition` with four rows, and then display this table. We can see that the second row has the empty string as its `ancestral_state`, and the third row's `ancestral_state` is `TTT`. Now let's print out the columns: ```{code-cell} ipython3 print("Ancestral state (numerical): ", s.ancestral_state) print("Ancestral state (as bytes): ", s.ancestral_state.tobytes()) print("Ancestral state offsets: ", s.ancestral_state_offset) ``` When we print out the tables `ancestral_state` column, we see that its a numpy array of length 5: this is the flattened array of [ASCII encoded](https://en.wikipedia.org/wiki/ASCII) values for these rows. When we decode these bytes using the numpy {meth}`tobytes` method, we get the string 'ATTTG'. This flattened array can now be transferred efficiently in memory like any other column We then use the `ancestral_state_offset` column to allow us find the individual rows. For a row `j`: ancestral_state[ancestral_state_offset[j]: ancestral_state_offset[j + 1]] gives us the array of bytes for the ancestral state in that row. For example, here is row 2: ```{code-cell} ipython3 s.ancestral_state[s.ancestral_state_offset[2]: s.ancestral_state_offset[3]].tobytes() ``` For a table with `n` rows, any offset column must have `n + 1` values, the first of which is always `0`. The values in this column must be nondecreasing, and cannot exceed the length of the ragged column in question. (sec_data_model_reference_sequence)= ## Reference sequence Along with the topology and site information stored in the tskit tree sequence, we can also optionally store an associated reference sequence. Reference sequences are flexible, and can consist simply of some metadata recording which assembly build a tree sequence uses, or storing the entire sequence itself. :::{warning} Reference sequence support in tskit is preliminary. Reference sequence data can be stored and accessed via the C API. Support in the Python API is limited to usage in {meth}`.TreeSequence.alignments` and related methods, where it provides the default values for nucleotide positions between {ref}`sites`. ::: (sec_data_model_tree_structure)= ## Tree structure (sec_data_model_quintuply_linked_trees)= ### Quintuply linked trees Tree structure in `tskit` is encoded internally as a "quintuply linked tree", a generalisation of the triply linked tree encoding used by Knuth and others. Nodes are represented by their integer IDs, and their relationships to other nodes are recorded in the `parent`, `left_child`, `right_child`, `left_sib` and `right_sib` arrays. For example, consider the following tree and its associated arrays: ```{code-cell} ipython3 :tags: ["hide-input"] import io import tskit from IPython.display import SVG nodes = """\ id is_sample time 0 1 0 1 1 0 2 1 0 3 1 0 4 1 0 5 0 1 6 0 2 7 0 3 """ edges = """\ left right parent child 0 60 5 4,3 0 40 6 2 0 60 6 1,0 20 40 6 5 0 20 7 5 40 60 7 5 0 60 7 6 40 60 7 2 """ ts = tskit.load_text( nodes=io.StringIO(nodes), edges=io.StringIO(edges), strict=False ) SVG(ts.first().draw_svg(time_scale="rank")) ``` ```{code-cell} ipython3 :tags: ["hide-input"] from IPython.display import HTML def html_quintuple_table(ts, show_virtual_root=False, show_convenience_arrays=False): tree = ts.first() columns = ["node", "parent", "left_child", "right_child", "left_sib", "right_sib"] convenience_arrays = ["num_children", "edge"] if show_convenience_arrays: columns += convenience_arrays data = {k:[] for k in columns} for u in sorted(tree.nodes(tree.virtual_root if show_virtual_root else None)): for colname in columns: data[colname].append(u if colname == "node" else getattr(tree, colname)(u)) html = "" for colname in columns: html += f"{colname}" html += "" for u in range(len(data["node"])): html += "" if u < ts.num_nodes else "" for colname in columns: html += f"{data[colname][u]}" html += "" return "" + html + "
" HTML(html_quintuple_table(ts)) ``` Each node in the tree corresponds to a row in this table, and the columns are the individual arrays recording the quintuply linked structure. Thus, we can see that the parent of nodes `0`, `1`, and `2` is `6`. Similarly, the left child of `6` is `0` and the right child of `6` is `2`. The `left_sib` and `right_sib` arrays then record each nodes sibling on its left or right, respectively; hence the right sib of `0` is `1`, and the right sib of `1` is `2`. Thus, sibling information allows us to efficiently support trees with arbitrary numbers of children. In each of the five pointer arrays, the null node (-1) is used to indicate the end of a path; thus, for example, the parent of `7` and left sib of `0` are null. Please see this {ref}`example ` for details of how to use the quintuply linked structure in the C API. :::{note} For many applications we do not need the quintuply linked trees, and (for example) the `left_sib` and `right_child` arrays can be ignored. The reason for using a quintuply instead of triply linked encoding is that it is not possible to efficiently update the trees as we move along the sequence without the quintuply linked structure. ::: :::{warning} The left-to-right ordering of nodes is determined by the order in which edges are inserted into the tree during iteration along the sequence. Thus, if we arrive at the same tree by iterating from different directions, the left-to-right ordering of nodes may be different! The specific ordering of the children of a node should therefore not be depended on. ::: ### Convenience arrays Similar to the five arrays representing the {ref}`quintuply linked tree`, convenience arrays track information on each node in the tree. These arrays are not essential to represent the trees within a tree sequence. However, they can be useful for specific algorithms (e.g. when computing tree (im)balance metrics). Two convenience arrays have been implemented so far: {attr}`Tree.num_children_array` and {attr}`Tree.edge_array`. Here is the table above with the convenience arrays also shown: ```{code-cell} ipython3 :tags: ["hide-input"] HTML(html_quintuple_table(ts, show_convenience_arrays=True)) ``` (sec_data_model_tree_roots)= ### Roots In the `tskit` {class}`trees ` we have shown so far, all the sample nodes have been connected to each other. This means each tree has only a single {attr}`~Tree.root` (i.e. the oldest node found when tracing a path backwards in time from any sample). However, a tree can contain {ref}`sec_data_model_tree_isolated_sample_nodes` or unconnected topologies, and can therefore have *multiple* {attr}`~Tree.roots`. Here's an example, created by deleting the edge joining `6` and `7` in the tree sequence used above: ```{code-cell} ipython3 :tags: ["hide-input"] tables = ts.dump_tables() tables.edges.truncate(ts.num_edges - 1) ts_multiroot = tables.tree_sequence() SVG(ts_multiroot.first().draw_svg(time_scale="rank")) ``` In `tskit` terminology, this should *not* be thought of as two separate trees, but as a single multi-root "tree", comprising two unlinked topologies. This fits with the definition of a tree in a tree sequence: a tree describes the ancestry of the same fixed set of sample nodes at a single position in the genome. In the picture above, *both* the left and right hand topologies are required to describe the genealogy of samples 0..4 at this position. Here's what the entire tree sequence now looks like: ```{code-cell} ipython3 :tags: ["hide-input"] SVG(ts_multiroot.draw_svg(time_scale="rank")) ``` From the terminology above, it can be seen that this tree sequence consists of only three trees (not five). The first tree, which applies from position 0 to 20, is the one used in our example. As we saw, removing the edge connecting node 6 to node 7 has created a tree with 2 roots (and thus 2 unconnected topologies in a single tree). In contrast, the second tree, from position 20 to 40, has a single root. Finally the third tree, from position 40 to 60, again has two roots. (sec_data_model_tree_root_threshold)= #### The root threshold The roots of a tree are defined by reference to the {ref}`sample nodes`. By default, roots are the unique endpoints of the paths traced upwards from the sample nodes; equivalently, each root counts one or more samples among its descendants (or is itself a sample node). This is the case when the {attr}`~Tree.root_threshold` property of a tree is left at its default value of `1`. If, however, the `root_threshold` is (say) `2`, then a node is considered a root only if it counts at least two samples among its descendants. Setting an alternative `root_threshold` value can be used to avoid visiting {ref}`sec_data_model_tree_isolated_sample_nodes`, for example when dealing with trees containing {ref}`sec_data_model_missing_data`. (sec_data_model_tree_virtual_root)= #### The virtual root To access all the {attr}`~Tree.roots` in a tree, tskit uses a special additional node called the **virtual root**. This is primarily a bookkeeping device, and can normally be ignored: it is not plotted in any visualizations and does not exist as an independent node in the node table. However, the virtual root can be useful in certain algorithms because its children are defined as all the "real" roots in a tree. Hence by descending downwards from the virtual root, it is possible to access the entire genealogy at a given site, even in a multi-root tree. In the quintuply linked tree encoding, the virtual root appears as an extra element at the end of each of the tree arrays. Here's the same table as before but with the virtual root also shown, using red italics to emphasise that it is not a "real" node: ```{code-cell} ipython3 :tags: ["hide-input"] HTML(html_quintuple_table(ts_multiroot, show_virtual_root=True)) ``` You can see that the virtual root (node 8) has 6 as its left child and 7 as its right child. Importantly, though, this is an asymmetric relationship: the parent of the "real" roots 6 and 7 is null (-1) and *not* the virtual root. Hence when we ascend up the tree from the sample nodes to their parents, we stop at the "real" roots, and never encounter the virtual root. Because the virtual root can be useful in some algorithms, it can optionally be returned in traversal orders (see {meth}`.Tree.nodes`). The following properties apply: - All trees in a tree sequence share the same virtual root. - The virtual root's ID is always equal to the number of nodes in the tree sequence (i.e. the length of the node table). However, there is **no corresponding row** in the node table, and any attempts to access information about the virtual root via either the tree sequence or tables APIs will fail with an out-of-bounds error. - The parent and siblings of the virtual root are null. - The time of the virtual root is defined as positive infinity (if accessed via {meth}`.Tree.time`). This is useful in defining the time-based node traversal orderings. - The virtual root is the parent of no other node---roots do **not** have parent pointers to the virtual root. (sec_data_model_tree_isolated_nodes)= ### Isolated nodes In a tree, it is possible for a node to have no children and no parent. Such a node is said to be *isolated*, meaning that we don't know anything about its relationships over a specific genomic interval. This is commonly true for ancestral genomes, which often have large regions that have not been inherited by any of the {ref}`sample nodes` in the tree sequence, and therefore regions about which we know nothing. This is true, for example, of node 7 in the middle tree of our previous example, which is why it is not plotted on that tree: ```{code-cell} ipython3 display(SVG(ts_multiroot.draw_svg(time_scale="rank"))) for tree in ts_multiroot.trees(): print( "Node 7", "is" if tree.is_isolated(7) else "is not", "isolated from position", tree.interval.left, "to", tree.interval.right, ) ``` (sec_data_model_tree_isolated_sample_nodes)= #### Isolated sample nodes It is also possible for a {ref}`sample node` to be isolated. As long as the {ref}`root threshold` is set to its default value, an isolated *sample* node will count as a root, and therefore be considered as being present on the tree (meaning it will be returned by the {meth}`Tree.nodes` and {meth}`Tree.samples` methods). When displaying a tree, isolated samples are shown unconnected to other nodes. To illustrate, we can remove the edge from node 2 to node 7: ```{code-cell} ipython3 :tags: ["hide-input"] tables = ts_multiroot.dump_tables() tables.edges.set_columns( **tables.edges[(tables.edges.parent != 7) | (tables.edges.child != 2)].asdict()) ts_isolated = tables.tree_sequence() SVG(ts_isolated.draw_svg(time_scale="rank")) ``` The rightmost tree now contains an isolated sample node (node 2), which counts as one of the {ref}`sec_data_model_tree_roots` of the tree. This tree therefore has three roots, one of which is node 2: ```{code-cell} ipython3 rightmost_tree = ts_isolated.at_index(-1) print(rightmost_tree.num_roots, "roots in the rightmost tree, with IDs", rightmost_tree.roots) print( "IDs of isolated samples in this tree:", [u for u in rightmost_tree.samples() if rightmost_tree.is_isolated(u)], ) ``` In `tskit`, isolated sample nodes are closely associated with the encoding of {ref}`sec_data_model_missing_data`. (sec_data_model_tree_dead_leaves_and_branches)= ### Dead leaves and branches In a `tskit` tree, a *leaf node* is defined as a node without any children. The implications of this turn out to be slighly unintuitive, and so are worth briefly documenting here. Firstly, the same node can be a leaf in one tree, and not a leaf in the next tree along the tree sequence. Secondly all isolated nodes must be leaves (as by definition they have no children). Thirdly sample nodes need not be leaves (they could be "internal samples"); likewise leaf nodes need not be samples. Node 7 in the example above provides a good case study. Note that it is a root node with at least one child (i.e. not a leaf) in trees 0 and 2; in contrast in tree 1 it is isolated. Strictly, because it is isolated in tree 1, it is also a leaf node there, although it is not attached to a root, not a sample, and is therefore not plotted. In this case, in that tree we can think of node 7 as a "dead leaf" (and we don't normally plot dead leaves). In fact, in a large tree sequence of many trees, most ancestral nodes will be isolated in any given tree, and therefore most nodes in such a tree will be of this sort. However, these dead leaves are excluded from most calculations on trees, because algorithms usually traverse the tree by starting at a root and working down, or by starting at a sample and working up. Hence when we refer to the leaves of a tree, it is usually shorthand for the leaves **on** the tree (that is, attached via branches, to one of the the tree roots). Dead leaves are excluded from this definition. Note that it is also possible to have trees in which there are "dead branches": that is sections of topology which are not accessible from a root, and whose tips are all dead leaves. Although valid, this is a relatively unusual state of affairs, and such branches are not plotted by the standard {ref}`sec_tskit_viz` methods. The {meth}`Tree.nodes` method will not, by default, traverse through dead branches, although it can be made to do so by specifying the ID of a dead node as the root for traversal. (sec_data_model_genetic_data)= ## Encoding genetic variation Genetic variation is incorporated into a tree sequence by placing {ref}`mutations` at {ref}`sites` along the genome. The genotypes of the different samples at each site can be found by using the tree to calculate which mutations are inherited by the different samples. This is the fundamental basis of how tree sequences efficiently encode DNA sequences, and is explained in depth elsewhere (e.g. {ref}`in the tutorials`). Below, we discuss some implications of this encoding in more detail, in particular the way in which it can be used to model missing data. (sec_data_model_missing_data)= ### Missing data If, at a particular genomic position, a node is {ref}`isolated` *and* additionally has no mutations directly above it, its genotype at that position is considered to be unknown (however, if there is a mutation above an isolated node, it can be thought of as saying directly what the genotype is, and so renders the genotype at that position not missing). By way of illustration, we'll use the {meth}`~TableCollection.delete_intervals` method to remove all knowledge of the ancestry in the middle portion of the previous example (say from position 15 to 45) sprinkle on some mutations, and make sure there are sites at every position: ```{code-cell} ipython3 :tags: ["hide-input"] import numpy as np import msprime tables = msprime.sim_mutations(ts_isolated, rate=0.1, random_seed=123).dump_tables() tables.delete_intervals([[15, 45]], simplify=False) missing_sites = np.setdiff1d(np.arange(tables.sequence_length), tables.sites.position) for pos in missing_sites: tables.sites.add_row(position=pos, ancestral_state="A") # Add sites at every pos tables.sort() missing_ts = tables.tree_sequence() SVG(missing_ts.draw_svg()) ``` The middle section of the genome now has no ancestry at all, and therefore for any site that is in this region, the genotypic state that it is assigned is a special value `tskit.MISSING_DATA`, or `-1`. The {meth}`~TreeSequence.haplotypes()` method, which outputs the actual allelic state for each sample, defaults to outputting an `N` at these sites. Therefore where any sample node is isolated, the haplotype will show an `N`, indicating the DNA sequence is unknown. This will be so not only in the middle of all of the sample genomes, but also at the right hand end of the genome of sample 2, as it is an isolated sample node in the rightmost tree: ```{code-cell} ipython3 for i, h in enumerate(missing_ts.haplotypes()): print(f"Sample {i}: {h}") ``` See the {meth}`TreeSequence.variants` method and {class}`Variant` class for more information on how missing data is represented in variant data. (sec_gotchas)= ## Possibly surprising consequences of the data model This is a section of miscellaneous issues that might trip even an experienced user up, also known as "gotchas". The current examples are quite uncommon, so can be ignored for most purposes, but the list may be expanded in the future. ### Unrelated material Usually, all parts of a tree sequence are ancestral to at least one sample, since that's essentially the definition of a sample: the genomes that we're describing the ancestry of. However, in some cases there will be portions of the tree sequence from which no samples inherit - notably, the result of a forwards simulation that has not been simplified. In fact, if the simulation has not coalesced, one can have entire portions of some marginal tree that are unrelated to any of the samples (for instance, an individual in the initial generation of the simulation that had no offspring). This can lead to a gotcha: the *roots* of a tree are defined to be only those roots *reachable from the samples* (and, furthermore, reachable from at least `root_threshold` samples; see {meth}`TreeSequence.trees`). So, our unlucky ancestor would not appear in the list of `roots`, even though if we drew all the relationships provided by the tree sequence, they'd definitely be a root. Furthermore, only nodes *reachable from a root* are included in the {meth}`Tree.nodes`. So, if you iterate over all the nodes in each marginal tree, you won't see those parts of the tree sequence that are unrelated to the samples. If you need to get those, too, you could either work with the {meth}`TreeSequence.edge_diffs` directly, or iterate over all nodes (instead of over {meth}`Tree.nodes`).